[developers] Re: unicode erg
danf at csli.stanford.edu
Sun May 15 16:56:41 CEST 2005
Hi Ben -
Thanks for the quick work on correcting the checked-in code even while
you're doing the spring cleaning. Once I made the corresponding name
changes in erg/lkb/script to use *lexdb-params* rather than the old
*psql-lexicon-parameters*, and in erg/lkb/mrs-initialization.lsp to use
*lexdb* rather than the old *psql-lexicon*, all now seems to work fine.
So I've checked in a corrected ERG lexicon, and will consult with my
local experts about the best way to make sure *locale* is set properly
for the ERG.
One more thing - in a relatively recent change, the method empty-cache()
for the lexdb now calls clear-generator-lexicon(), but this is not ideal,
since empty-cache() is called via clear-expanded-lex() and its caller
uncache-lexicon() by the function tsdb::finalize-run() at the end of
each batch run using the tsdb++ machinery. I'd be grateful if you could
revisit the call in empty-cache() so the generator index can survive
beyond a single batch run.
More information about the developers